{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "global_step": 5930,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 9.995951417004049e-06,
      "loss": 4.0911,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.991902834008098e-06,
      "loss": 2.8012,
      "step": 20
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.987854251012146e-06,
      "loss": 2.389,
      "step": 30
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.983805668016194e-06,
      "loss": 2.1658,
      "step": 40
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.979757085020244e-06,
      "loss": 2.026,
      "step": 50
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.975708502024292e-06,
      "loss": 1.7846,
      "step": 60
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.971659919028341e-06,
      "loss": 1.703,
      "step": 70
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.96761133603239e-06,
      "loss": 1.5691,
      "step": 80
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.963562753036437e-06,
      "loss": 1.4297,
      "step": 90
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.959514170040487e-06,
      "loss": 1.4341,
      "step": 100
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.955465587044535e-06,
      "loss": 1.3304,
      "step": 110
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.951417004048583e-06,
      "loss": 1.3988,
      "step": 120
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.947368421052632e-06,
      "loss": 1.3272,
      "step": 130
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.943319838056682e-06,
      "loss": 1.2999,
      "step": 140
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.93927125506073e-06,
      "loss": 1.2997,
      "step": 150
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.935222672064778e-06,
      "loss": 1.3704,
      "step": 160
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.931174089068828e-06,
      "loss": 1.3313,
      "step": 170
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.927125506072876e-06,
      "loss": 1.3554,
      "step": 180
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.923076923076923e-06,
      "loss": 1.41,
      "step": 190
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.919028340080973e-06,
      "loss": 1.3005,
      "step": 200
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.914979757085021e-06,
      "loss": 1.2311,
      "step": 210
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.910931174089069e-06,
      "loss": 1.3162,
      "step": 220
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.906882591093119e-06,
      "loss": 1.2895,
      "step": 230
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.902834008097167e-06,
      "loss": 1.2812,
      "step": 240
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.898785425101216e-06,
      "loss": 1.2682,
      "step": 250
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.894736842105264e-06,
      "loss": 1.2992,
      "step": 260
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.890688259109312e-06,
      "loss": 1.2832,
      "step": 270
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.886639676113362e-06,
      "loss": 1.2912,
      "step": 280
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.88259109311741e-06,
      "loss": 1.2949,
      "step": 290
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.878542510121458e-06,
      "loss": 1.2557,
      "step": 300
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.874493927125507e-06,
      "loss": 1.3119,
      "step": 310
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.870445344129555e-06,
      "loss": 1.3092,
      "step": 320
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.866396761133603e-06,
      "loss": 1.2441,
      "step": 330
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.862348178137653e-06,
      "loss": 1.3288,
      "step": 340
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.8582995951417e-06,
      "loss": 1.2382,
      "step": 350
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.85425101214575e-06,
      "loss": 1.2819,
      "step": 360
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.850202429149798e-06,
      "loss": 1.3011,
      "step": 370
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.846153846153846e-06,
      "loss": 1.2421,
      "step": 380
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.842105263157896e-06,
      "loss": 1.2071,
      "step": 390
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.838056680161945e-06,
      "loss": 1.2838,
      "step": 400
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.834008097165993e-06,
      "loss": 1.298,
      "step": 410
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.829959514170041e-06,
      "loss": 1.2867,
      "step": 420
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.825910931174091e-06,
      "loss": 1.1868,
      "step": 430
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.821862348178139e-06,
      "loss": 1.2263,
      "step": 440
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.817813765182187e-06,
      "loss": 1.2701,
      "step": 450
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.813765182186236e-06,
      "loss": 1.2542,
      "step": 460
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.809716599190284e-06,
      "loss": 1.2833,
      "step": 470
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.805668016194332e-06,
      "loss": 1.2109,
      "step": 480
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.801619433198382e-06,
      "loss": 1.2657,
      "step": 490
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.79757085020243e-06,
      "loss": 1.2587,
      "step": 500
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.79352226720648e-06,
      "loss": 1.2647,
      "step": 510
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.789473684210527e-06,
      "loss": 1.1917,
      "step": 520
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.785425101214575e-06,
      "loss": 1.2281,
      "step": 530
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.781376518218625e-06,
      "loss": 1.2677,
      "step": 540
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.777327935222673e-06,
      "loss": 1.2299,
      "step": 550
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.77327935222672e-06,
      "loss": 1.2524,
      "step": 560
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.76923076923077e-06,
      "loss": 1.2902,
      "step": 570
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.765182186234818e-06,
      "loss": 1.2603,
      "step": 580
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.761133603238866e-06,
      "loss": 1.2572,
      "step": 590
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.757085020242916e-06,
      "loss": 1.2223,
      "step": 600
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.753036437246964e-06,
      "loss": 1.2284,
      "step": 610
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.748987854251012e-06,
      "loss": 1.2331,
      "step": 620
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.744939271255061e-06,
      "loss": 1.2601,
      "step": 630
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.74089068825911e-06,
      "loss": 1.195,
      "step": 640
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.736842105263159e-06,
      "loss": 1.2176,
      "step": 650
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.732793522267207e-06,
      "loss": 1.2684,
      "step": 660
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.728744939271257e-06,
      "loss": 1.2575,
      "step": 670
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.724696356275305e-06,
      "loss": 1.2187,
      "step": 680
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.720647773279354e-06,
      "loss": 1.2184,
      "step": 690
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.716599190283402e-06,
      "loss": 1.282,
      "step": 700
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.71255060728745e-06,
      "loss": 1.2464,
      "step": 710
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.7085020242915e-06,
      "loss": 1.2041,
      "step": 720
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.704453441295548e-06,
      "loss": 1.26,
      "step": 730
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.700404858299596e-06,
      "loss": 1.2401,
      "step": 740
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.696356275303645e-06,
      "loss": 1.2289,
      "step": 750
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.692307692307693e-06,
      "loss": 1.2587,
      "step": 760
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.688259109311741e-06,
      "loss": 1.2078,
      "step": 770
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.68421052631579e-06,
      "loss": 1.192,
      "step": 780
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.680161943319839e-06,
      "loss": 1.2981,
      "step": 790
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.676113360323888e-06,
      "loss": 1.1942,
      "step": 800
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.672064777327936e-06,
      "loss": 1.247,
      "step": 810
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.668016194331984e-06,
      "loss": 1.1902,
      "step": 820
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.663967611336034e-06,
      "loss": 1.2271,
      "step": 830
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.659919028340082e-06,
      "loss": 1.2407,
      "step": 840
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.65587044534413e-06,
      "loss": 1.2067,
      "step": 850
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.65182186234818e-06,
      "loss": 1.2284,
      "step": 860
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.647773279352227e-06,
      "loss": 1.2676,
      "step": 870
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.643724696356275e-06,
      "loss": 1.2542,
      "step": 880
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.639676113360325e-06,
      "loss": 1.2364,
      "step": 890
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.635627530364373e-06,
      "loss": 1.3608,
      "step": 900
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.63157894736842e-06,
      "loss": 1.2739,
      "step": 910
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.62753036437247e-06,
      "loss": 1.1984,
      "step": 920
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.62348178137652e-06,
      "loss": 1.1875,
      "step": 930
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.619433198380568e-06,
      "loss": 1.2447,
      "step": 940
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.615384615384616e-06,
      "loss": 1.26,
      "step": 950
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.611336032388665e-06,
      "loss": 1.2059,
      "step": 960
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.607287449392713e-06,
      "loss": 1.233,
      "step": 970
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.603238866396763e-06,
      "loss": 1.244,
      "step": 980
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.59919028340081e-06,
      "loss": 1.2613,
      "step": 990
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.595141700404859e-06,
      "loss": 1.2274,
      "step": 1000
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.591093117408908e-06,
      "loss": 1.2727,
      "step": 1010
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.587044534412956e-06,
      "loss": 1.2834,
      "step": 1020
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.582995951417004e-06,
      "loss": 1.259,
      "step": 1030
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.578947368421054e-06,
      "loss": 1.2742,
      "step": 1040
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.574898785425102e-06,
      "loss": 1.2658,
      "step": 1050
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.57085020242915e-06,
      "loss": 1.2203,
      "step": 1060
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.5668016194332e-06,
      "loss": 1.1757,
      "step": 1070
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.562753036437247e-06,
      "loss": 1.2296,
      "step": 1080
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.558704453441297e-06,
      "loss": 1.2714,
      "step": 1090
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.554655870445345e-06,
      "loss": 1.2265,
      "step": 1100
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.550607287449393e-06,
      "loss": 1.1984,
      "step": 1110
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.546558704453442e-06,
      "loss": 1.2706,
      "step": 1120
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.54251012145749e-06,
      "loss": 1.2611,
      "step": 1130
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.538461538461538e-06,
      "loss": 1.2584,
      "step": 1140
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.534412955465588e-06,
      "loss": 1.2849,
      "step": 1150
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.530364372469636e-06,
      "loss": 1.2254,
      "step": 1160
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.526315789473686e-06,
      "loss": 1.2698,
      "step": 1170
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.522267206477733e-06,
      "loss": 1.2037,
      "step": 1180
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.518218623481783e-06,
      "loss": 1.2408,
      "step": 1190
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.514170040485831e-06,
      "loss": 1.2116,
      "step": 1200
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.510121457489879e-06,
      "loss": 1.2393,
      "step": 1210
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.506072874493929e-06,
      "loss": 1.2177,
      "step": 1220
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.502024291497977e-06,
      "loss": 1.23,
      "step": 1230
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.497975708502026e-06,
      "loss": 1.2245,
      "step": 1240
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.493927125506074e-06,
      "loss": 1.2602,
      "step": 1250
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.489878542510122e-06,
      "loss": 1.2017,
      "step": 1260
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.485829959514172e-06,
      "loss": 1.2281,
      "step": 1270
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.48178137651822e-06,
      "loss": 1.2808,
      "step": 1280
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.477732793522268e-06,
      "loss": 1.2017,
      "step": 1290
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.473684210526317e-06,
      "loss": 1.2557,
      "step": 1300
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.469635627530365e-06,
      "loss": 1.2952,
      "step": 1310
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.465587044534413e-06,
      "loss": 1.2477,
      "step": 1320
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.461538461538463e-06,
      "loss": 1.2397,
      "step": 1330
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.45748987854251e-06,
      "loss": 1.1556,
      "step": 1340
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.453441295546559e-06,
      "loss": 1.217,
      "step": 1350
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.449392712550608e-06,
      "loss": 1.2725,
      "step": 1360
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.445344129554656e-06,
      "loss": 1.2079,
      "step": 1370
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.441295546558706e-06,
      "loss": 1.2646,
      "step": 1380
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.437246963562754e-06,
      "loss": 1.2034,
      "step": 1390
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.433198380566802e-06,
      "loss": 1.2442,
      "step": 1400
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.429149797570851e-06,
      "loss": 1.2463,
      "step": 1410
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.4251012145749e-06,
      "loss": 1.2257,
      "step": 1420
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.421052631578947e-06,
      "loss": 1.2969,
      "step": 1430
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.417004048582997e-06,
      "loss": 1.2288,
      "step": 1440
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.412955465587045e-06,
      "loss": 1.2307,
      "step": 1450
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.408906882591094e-06,
      "loss": 1.2101,
      "step": 1460
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.404858299595142e-06,
      "loss": 1.2009,
      "step": 1470
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.400809716599192e-06,
      "loss": 1.2698,
      "step": 1480
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.1532880067825317,
      "eval_runtime": 11.4194,
      "eval_samples_per_second": 96.327,
      "eval_steps_per_second": 12.085,
      "step": 1482
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.39676113360324e-06,
      "loss": 1.1808,
      "step": 1490
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.392712550607288e-06,
      "loss": 1.1629,
      "step": 1500
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.388663967611337e-06,
      "loss": 1.1656,
      "step": 1510
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.384615384615385e-06,
      "loss": 1.1163,
      "step": 1520
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.380566801619435e-06,
      "loss": 1.1403,
      "step": 1530
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.376518218623483e-06,
      "loss": 1.122,
      "step": 1540
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.37246963562753e-06,
      "loss": 1.1426,
      "step": 1550
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.36842105263158e-06,
      "loss": 1.1142,
      "step": 1560
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.364372469635628e-06,
      "loss": 1.1143,
      "step": 1570
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.360323886639676e-06,
      "loss": 1.1798,
      "step": 1580
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.356275303643726e-06,
      "loss": 1.1575,
      "step": 1590
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.352226720647774e-06,
      "loss": 1.1459,
      "step": 1600
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.348178137651822e-06,
      "loss": 1.199,
      "step": 1610
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.344129554655871e-06,
      "loss": 1.1568,
      "step": 1620
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.34008097165992e-06,
      "loss": 1.1577,
      "step": 1630
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.336032388663967e-06,
      "loss": 1.0549,
      "step": 1640
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.331983805668017e-06,
      "loss": 1.1889,
      "step": 1650
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.327935222672065e-06,
      "loss": 1.1355,
      "step": 1660
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.323886639676115e-06,
      "loss": 1.1725,
      "step": 1670
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.319838056680162e-06,
      "loss": 1.1096,
      "step": 1680
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.315789473684212e-06,
      "loss": 1.2392,
      "step": 1690
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.31174089068826e-06,
      "loss": 1.1508,
      "step": 1700
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.30769230769231e-06,
      "loss": 1.1437,
      "step": 1710
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.303643724696358e-06,
      "loss": 1.204,
      "step": 1720
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.299595141700406e-06,
      "loss": 1.1795,
      "step": 1730
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.295546558704455e-06,
      "loss": 1.112,
      "step": 1740
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.291497975708503e-06,
      "loss": 1.2072,
      "step": 1750
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.287449392712551e-06,
      "loss": 1.1518,
      "step": 1760
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.2834008097166e-06,
      "loss": 1.1479,
      "step": 1770
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.279352226720649e-06,
      "loss": 1.1183,
      "step": 1780
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.275303643724697e-06,
      "loss": 1.1493,
      "step": 1790
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.271255060728746e-06,
      "loss": 1.1603,
      "step": 1800
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.267206477732794e-06,
      "loss": 1.186,
      "step": 1810
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.263157894736844e-06,
      "loss": 1.1857,
      "step": 1820
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.259109311740892e-06,
      "loss": 1.1957,
      "step": 1830
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.25506072874494e-06,
      "loss": 1.1102,
      "step": 1840
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.25101214574899e-06,
      "loss": 1.1439,
      "step": 1850
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.246963562753037e-06,
      "loss": 1.1252,
      "step": 1860
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.242914979757085e-06,
      "loss": 1.1238,
      "step": 1870
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.238866396761135e-06,
      "loss": 1.135,
      "step": 1880
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.234817813765183e-06,
      "loss": 1.1542,
      "step": 1890
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.23076923076923e-06,
      "loss": 1.1481,
      "step": 1900
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.22672064777328e-06,
      "loss": 1.0846,
      "step": 1910
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.222672064777328e-06,
      "loss": 1.1346,
      "step": 1920
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.218623481781376e-06,
      "loss": 1.0831,
      "step": 1930
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.214574898785426e-06,
      "loss": 1.2031,
      "step": 1940
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.210526315789474e-06,
      "loss": 1.0651,
      "step": 1950
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.206477732793523e-06,
      "loss": 1.1237,
      "step": 1960
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.202429149797571e-06,
      "loss": 1.0612,
      "step": 1970
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.198380566801621e-06,
      "loss": 1.1377,
      "step": 1980
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.194331983805669e-06,
      "loss": 1.1692,
      "step": 1990
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.190283400809718e-06,
      "loss": 1.1136,
      "step": 2000
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.186234817813766e-06,
      "loss": 1.1361,
      "step": 2010
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.182186234817814e-06,
      "loss": 1.1663,
      "step": 2020
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.178137651821864e-06,
      "loss": 1.1054,
      "step": 2030
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.174089068825912e-06,
      "loss": 1.1427,
      "step": 2040
    },
    {
      "epoch": 1.38,
      "learning_rate": 9.17004048582996e-06,
      "loss": 1.1392,
      "step": 2050
    },
    {
      "epoch": 1.39,
      "learning_rate": 9.16599190283401e-06,
      "loss": 1.191,
      "step": 2060
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.161943319838057e-06,
      "loss": 1.1006,
      "step": 2070
    },
    {
      "epoch": 1.4,
      "learning_rate": 9.157894736842105e-06,
      "loss": 1.1372,
      "step": 2080
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.153846153846155e-06,
      "loss": 1.1428,
      "step": 2090
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.149797570850203e-06,
      "loss": 1.1433,
      "step": 2100
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.145748987854253e-06,
      "loss": 1.0997,
      "step": 2110
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.1417004048583e-06,
      "loss": 1.1478,
      "step": 2120
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.137651821862348e-06,
      "loss": 1.0928,
      "step": 2130
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.133603238866398e-06,
      "loss": 1.0993,
      "step": 2140
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.129554655870446e-06,
      "loss": 1.1306,
      "step": 2150
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.125506072874494e-06,
      "loss": 1.1527,
      "step": 2160
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.121457489878544e-06,
      "loss": 1.1394,
      "step": 2170
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.117408906882591e-06,
      "loss": 1.1787,
      "step": 2180
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.11336032388664e-06,
      "loss": 1.1295,
      "step": 2190
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.109311740890689e-06,
      "loss": 1.1843,
      "step": 2200
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.105263157894739e-06,
      "loss": 1.1243,
      "step": 2210
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.101214574898785e-06,
      "loss": 1.1748,
      "step": 2220
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.097165991902835e-06,
      "loss": 1.1282,
      "step": 2230
    },
    {
      "epoch": 1.51,
      "learning_rate": 9.093117408906884e-06,
      "loss": 1.0969,
      "step": 2240
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.089068825910932e-06,
      "loss": 1.158,
      "step": 2250
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.085020242914982e-06,
      "loss": 1.1417,
      "step": 2260
    },
    {
      "epoch": 1.53,
      "learning_rate": 9.08097165991903e-06,
      "loss": 1.1397,
      "step": 2270
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.076923076923078e-06,
      "loss": 1.1622,
      "step": 2280
    },
    {
      "epoch": 1.54,
      "learning_rate": 9.072874493927127e-06,
      "loss": 1.1217,
      "step": 2290
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.068825910931175e-06,
      "loss": 1.1527,
      "step": 2300
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.064777327935223e-06,
      "loss": 1.1927,
      "step": 2310
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.060728744939273e-06,
      "loss": 1.1505,
      "step": 2320
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.05668016194332e-06,
      "loss": 1.123,
      "step": 2330
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.052631578947369e-06,
      "loss": 1.1453,
      "step": 2340
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.048582995951418e-06,
      "loss": 1.1903,
      "step": 2350
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.044534412955466e-06,
      "loss": 1.144,
      "step": 2360
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.040485829959514e-06,
      "loss": 1.1469,
      "step": 2370
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.036437246963564e-06,
      "loss": 1.1463,
      "step": 2380
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.032388663967612e-06,
      "loss": 1.1702,
      "step": 2390
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.028340080971661e-06,
      "loss": 1.078,
      "step": 2400
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.02429149797571e-06,
      "loss": 1.0907,
      "step": 2410
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.020242914979757e-06,
      "loss": 1.1621,
      "step": 2420
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.016194331983807e-06,
      "loss": 1.0947,
      "step": 2430
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.012145748987855e-06,
      "loss": 1.1301,
      "step": 2440
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.008097165991903e-06,
      "loss": 1.1232,
      "step": 2450
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.004048582995952e-06,
      "loss": 1.1737,
      "step": 2460
    },
    {
      "epoch": 1.67,
      "learning_rate": 9e-06,
      "loss": 1.1475,
      "step": 2470
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.99595141700405e-06,
      "loss": 1.1242,
      "step": 2480
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.991902834008098e-06,
      "loss": 1.1321,
      "step": 2490
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.987854251012147e-06,
      "loss": 1.1131,
      "step": 2500
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.983805668016195e-06,
      "loss": 1.1456,
      "step": 2510
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.979757085020243e-06,
      "loss": 1.0983,
      "step": 2520
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.975708502024293e-06,
      "loss": 1.1598,
      "step": 2530
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.971659919028341e-06,
      "loss": 1.1861,
      "step": 2540
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.96761133603239e-06,
      "loss": 1.1334,
      "step": 2550
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.963562753036438e-06,
      "loss": 1.1605,
      "step": 2560
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.959514170040486e-06,
      "loss": 1.136,
      "step": 2570
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.955465587044536e-06,
      "loss": 1.1162,
      "step": 2580
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.951417004048584e-06,
      "loss": 1.1115,
      "step": 2590
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.947368421052632e-06,
      "loss": 1.1025,
      "step": 2600
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.943319838056681e-06,
      "loss": 1.1616,
      "step": 2610
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.93927125506073e-06,
      "loss": 1.1407,
      "step": 2620
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.935222672064777e-06,
      "loss": 1.1016,
      "step": 2630
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.931174089068827e-06,
      "loss": 1.111,
      "step": 2640
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.927125506072875e-06,
      "loss": 1.1226,
      "step": 2650
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.923076923076923e-06,
      "loss": 1.1105,
      "step": 2660
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.919028340080972e-06,
      "loss": 1.1227,
      "step": 2670
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.91497975708502e-06,
      "loss": 1.1462,
      "step": 2680
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.91093117408907e-06,
      "loss": 1.1119,
      "step": 2690
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.906882591093118e-06,
      "loss": 1.1567,
      "step": 2700
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.902834008097166e-06,
      "loss": 1.1565,
      "step": 2710
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.898785425101216e-06,
      "loss": 1.1116,
      "step": 2720
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.894736842105265e-06,
      "loss": 1.1139,
      "step": 2730
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.890688259109311e-06,
      "loss": 1.1454,
      "step": 2740
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.886639676113361e-06,
      "loss": 1.1084,
      "step": 2750
    },
    {
      "epoch": 1.86,
      "learning_rate": 8.882591093117409e-06,
      "loss": 1.1829,
      "step": 2760
    },
    {
      "epoch": 1.87,
      "learning_rate": 8.878542510121459e-06,
      "loss": 1.1262,
      "step": 2770
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.874493927125507e-06,
      "loss": 1.1754,
      "step": 2780
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.870445344129556e-06,
      "loss": 1.1269,
      "step": 2790
    },
    {
      "epoch": 1.89,
      "learning_rate": 8.866396761133604e-06,
      "loss": 1.1663,
      "step": 2800
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.862348178137652e-06,
      "loss": 1.1457,
      "step": 2810
    },
    {
      "epoch": 1.9,
      "learning_rate": 8.858299595141702e-06,
      "loss": 1.1143,
      "step": 2820
    },
    {
      "epoch": 1.91,
      "learning_rate": 8.85425101214575e-06,
      "loss": 1.1956,
      "step": 2830
    },
    {
      "epoch": 1.92,
      "learning_rate": 8.8502024291498e-06,
      "loss": 1.152,
      "step": 2840
    },
    {
      "epoch": 1.92,
      "learning_rate": 8.846153846153847e-06,
      "loss": 1.1365,
      "step": 2850
    },
    {
      "epoch": 1.93,
      "learning_rate": 8.842105263157895e-06,
      "loss": 1.1519,
      "step": 2860
    },
    {
      "epoch": 1.94,
      "learning_rate": 8.838056680161945e-06,
      "loss": 1.126,
      "step": 2870
    },
    {
      "epoch": 1.94,
      "learning_rate": 8.834008097165993e-06,
      "loss": 1.1408,
      "step": 2880
    },
    {
      "epoch": 1.95,
      "learning_rate": 8.82995951417004e-06,
      "loss": 1.1522,
      "step": 2890
    },
    {
      "epoch": 1.96,
      "learning_rate": 8.82591093117409e-06,
      "loss": 1.1236,
      "step": 2900
    },
    {
      "epoch": 1.96,
      "learning_rate": 8.821862348178138e-06,
      "loss": 1.0927,
      "step": 2910
    },
    {
      "epoch": 1.97,
      "learning_rate": 8.817813765182186e-06,
      "loss": 1.21,
      "step": 2920
    },
    {
      "epoch": 1.98,
      "learning_rate": 8.813765182186236e-06,
      "loss": 1.1119,
      "step": 2930
    },
    {
      "epoch": 1.98,
      "learning_rate": 8.809716599190284e-06,
      "loss": 1.1383,
      "step": 2940
    },
    {
      "epoch": 1.99,
      "learning_rate": 8.805668016194333e-06,
      "loss": 1.1438,
      "step": 2950
    },
    {
      "epoch": 2.0,
      "learning_rate": 8.801619433198381e-06,
      "loss": 1.1747,
      "step": 2960
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.1016342639923096,
      "eval_runtime": 10.6294,
      "eval_samples_per_second": 103.487,
      "eval_steps_per_second": 12.983,
      "step": 2965
    },
    {
      "epoch": 2.0,
      "learning_rate": 8.79757085020243e-06,
      "loss": 1.083,
      "step": 2970
    },
    {
      "epoch": 2.01,
      "learning_rate": 8.793522267206479e-06,
      "loss": 1.0542,
      "step": 2980
    },
    {
      "epoch": 2.02,
      "learning_rate": 8.789473684210527e-06,
      "loss": 1.0974,
      "step": 2990
    },
    {
      "epoch": 2.02,
      "learning_rate": 8.785425101214576e-06,
      "loss": 1.0074,
      "step": 3000
    },
    {
      "epoch": 2.03,
      "learning_rate": 8.781376518218624e-06,
      "loss": 1.0825,
      "step": 3010
    },
    {
      "epoch": 2.04,
      "learning_rate": 8.777327935222674e-06,
      "loss": 1.065,
      "step": 3020
    },
    {
      "epoch": 2.04,
      "learning_rate": 8.77327935222672e-06,
      "loss": 1.0996,
      "step": 3030
    },
    {
      "epoch": 2.05,
      "learning_rate": 8.76923076923077e-06,
      "loss": 1.1079,
      "step": 3040
    },
    {
      "epoch": 2.06,
      "learning_rate": 8.76518218623482e-06,
      "loss": 1.0803,
      "step": 3050
    },
    {
      "epoch": 2.06,
      "learning_rate": 8.761133603238867e-06,
      "loss": 1.0923,
      "step": 3060
    },
    {
      "epoch": 2.07,
      "learning_rate": 8.757085020242915e-06,
      "loss": 1.1171,
      "step": 3070
    },
    {
      "epoch": 2.08,
      "learning_rate": 8.753036437246965e-06,
      "loss": 1.0404,
      "step": 3080
    },
    {
      "epoch": 2.08,
      "learning_rate": 8.748987854251013e-06,
      "loss": 1.0881,
      "step": 3090
    },
    {
      "epoch": 2.09,
      "learning_rate": 8.744939271255061e-06,
      "loss": 1.1536,
      "step": 3100
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.74089068825911e-06,
      "loss": 1.0868,
      "step": 3110
    },
    {
      "epoch": 2.1,
      "learning_rate": 8.736842105263158e-06,
      "loss": 1.1422,
      "step": 3120
    },
    {
      "epoch": 2.11,
      "learning_rate": 8.732793522267208e-06,
      "loss": 1.1048,
      "step": 3130
    },
    {
      "epoch": 2.12,
      "learning_rate": 8.728744939271256e-06,
      "loss": 1.136,
      "step": 3140
    },
    {
      "epoch": 2.12,
      "learning_rate": 8.724696356275304e-06,
      "loss": 1.07,
      "step": 3150
    },
    {
      "epoch": 2.13,
      "learning_rate": 8.720647773279354e-06,
      "loss": 1.1336,
      "step": 3160
    },
    {
      "epoch": 2.14,
      "learning_rate": 8.716599190283401e-06,
      "loss": 1.0946,
      "step": 3170
    },
    {
      "epoch": 2.15,
      "learning_rate": 8.71255060728745e-06,
      "loss": 1.061,
      "step": 3180
    },
    {
      "epoch": 2.15,
      "learning_rate": 8.708502024291499e-06,
      "loss": 1.0275,
      "step": 3190
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.704453441295547e-06,
      "loss": 1.0665,
      "step": 3200
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.700404858299595e-06,
      "loss": 1.0563,
      "step": 3210
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.696356275303645e-06,
      "loss": 1.0756,
      "step": 3220
    },
    {
      "epoch": 2.18,
      "learning_rate": 8.692307692307692e-06,
      "loss": 1.0981,
      "step": 3230
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.688259109311742e-06,
      "loss": 1.1138,
      "step": 3240
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.68421052631579e-06,
      "loss": 1.0838,
      "step": 3250
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.680161943319838e-06,
      "loss": 1.1053,
      "step": 3260
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.676113360323888e-06,
      "loss": 1.1099,
      "step": 3270
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.672064777327936e-06,
      "loss": 1.0884,
      "step": 3280
    },
    {
      "epoch": 2.22,
      "learning_rate": 8.668016194331985e-06,
      "loss": 1.0857,
      "step": 3290
    },
    {
      "epoch": 2.23,
      "learning_rate": 8.663967611336033e-06,
      "loss": 1.0629,
      "step": 3300
    },
    {
      "epoch": 2.23,
      "learning_rate": 8.659919028340083e-06,
      "loss": 1.0753,
      "step": 3310
    },
    {
      "epoch": 2.24,
      "learning_rate": 8.65587044534413e-06,
      "loss": 1.1481,
      "step": 3320
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.651821862348179e-06,
      "loss": 1.1127,
      "step": 3330
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.647773279352228e-06,
      "loss": 1.1666,
      "step": 3340
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.643724696356276e-06,
      "loss": 1.1086,
      "step": 3350
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.639676113360324e-06,
      "loss": 1.1102,
      "step": 3360
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.635627530364374e-06,
      "loss": 1.1194,
      "step": 3370
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.631578947368422e-06,
      "loss": 1.145,
      "step": 3380
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.62753036437247e-06,
      "loss": 1.1103,
      "step": 3390
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.62348178137652e-06,
      "loss": 1.0998,
      "step": 3400
    },
    {
      "epoch": 2.3,
      "learning_rate": 8.619433198380567e-06,
      "loss": 1.1551,
      "step": 3410
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.615384615384615e-06,
      "loss": 1.0687,
      "step": 3420
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.611336032388665e-06,
      "loss": 1.0959,
      "step": 3430
    },
    {
      "epoch": 2.32,
      "learning_rate": 8.607287449392713e-06,
      "loss": 1.0799,
      "step": 3440
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.60323886639676e-06,
      "loss": 1.0673,
      "step": 3450
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.59919028340081e-06,
      "loss": 1.0873,
      "step": 3460
    },
    {
      "epoch": 2.34,
      "learning_rate": 8.595141700404858e-06,
      "loss": 1.1119,
      "step": 3470
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.591093117408906e-06,
      "loss": 1.1356,
      "step": 3480
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.587044534412956e-06,
      "loss": 1.0924,
      "step": 3490
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.582995951417005e-06,
      "loss": 1.0818,
      "step": 3500
    },
    {
      "epoch": 2.37,
      "learning_rate": 8.578947368421053e-06,
      "loss": 1.1007,
      "step": 3510
    },
    {
      "epoch": 2.37,
      "learning_rate": 8.574898785425103e-06,
      "loss": 1.0591,
      "step": 3520
    },
    {
      "epoch": 2.38,
      "learning_rate": 8.57085020242915e-06,
      "loss": 1.1069,
      "step": 3530
    },
    {
      "epoch": 2.39,
      "learning_rate": 8.566801619433199e-06,
      "loss": 1.1191,
      "step": 3540
    },
    {
      "epoch": 2.39,
      "learning_rate": 8.562753036437247e-06,
      "loss": 1.077,
      "step": 3550
    },
    {
      "epoch": 2.4,
      "learning_rate": 8.558704453441296e-06,
      "loss": 1.104,
      "step": 3560
    },
    {
      "epoch": 2.41,
      "learning_rate": 8.554655870445344e-06,
      "loss": 1.0758,
      "step": 3570
    },
    {
      "epoch": 2.41,
      "learning_rate": 8.550607287449394e-06,
      "loss": 1.0885,
      "step": 3580
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.546558704453442e-06,
      "loss": 1.084,
      "step": 3590
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.54251012145749e-06,
      "loss": 1.0869,
      "step": 3600
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.53846153846154e-06,
      "loss": 1.073,
      "step": 3610
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.534412955465587e-06,
      "loss": 1.0766,
      "step": 3620
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.530364372469635e-06,
      "loss": 1.1085,
      "step": 3630
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.526315789473685e-06,
      "loss": 1.097,
      "step": 3640
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.522267206477733e-06,
      "loss": 1.0968,
      "step": 3650
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.518218623481783e-06,
      "loss": 1.0917,
      "step": 3660
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.51417004048583e-06,
      "loss": 1.047,
      "step": 3670
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.510121457489878e-06,
      "loss": 1.0469,
      "step": 3680
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.506072874493928e-06,
      "loss": 1.1114,
      "step": 3690
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.502024291497976e-06,
      "loss": 1.0319,
      "step": 3700
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.497975708502024e-06,
      "loss": 1.0733,
      "step": 3710
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.493927125506074e-06,
      "loss": 1.1237,
      "step": 3720
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.489878542510121e-06,
      "loss": 1.0649,
      "step": 3730
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.48582995951417e-06,
      "loss": 1.0903,
      "step": 3740
    },
    {
      "epoch": 2.53,
      "learning_rate": 8.481781376518219e-06,
      "loss": 1.0633,
      "step": 3750
    },
    {
      "epoch": 2.54,
      "learning_rate": 8.477732793522267e-06,
      "loss": 1.1259,
      "step": 3760
    },
    {
      "epoch": 2.54,
      "learning_rate": 8.473684210526317e-06,
      "loss": 1.0978,
      "step": 3770
    },
    {
      "epoch": 2.55,
      "learning_rate": 8.469635627530365e-06,
      "loss": 1.1362,
      "step": 3780
    },
    {
      "epoch": 2.56,
      "learning_rate": 8.465587044534414e-06,
      "loss": 1.1487,
      "step": 3790
    },
    {
      "epoch": 2.56,
      "learning_rate": 8.461538461538462e-06,
      "loss": 1.0887,
      "step": 3800
    },
    {
      "epoch": 2.57,
      "learning_rate": 8.457489878542512e-06,
      "loss": 1.1039,
      "step": 3810
    },
    {
      "epoch": 2.58,
      "learning_rate": 8.453441295546558e-06,
      "loss": 1.0906,
      "step": 3820
    },
    {
      "epoch": 2.58,
      "learning_rate": 8.449392712550608e-06,
      "loss": 1.1039,
      "step": 3830
    },
    {
      "epoch": 2.59,
      "learning_rate": 8.445344129554657e-06,
      "loss": 1.0432,
      "step": 3840
    },
    {
      "epoch": 2.6,
      "learning_rate": 8.441295546558705e-06,
      "loss": 1.1565,
      "step": 3850
    },
    {
      "epoch": 2.6,
      "learning_rate": 8.437246963562753e-06,
      "loss": 1.1786,
      "step": 3860
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.433198380566803e-06,
      "loss": 1.0724,
      "step": 3870
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.42914979757085e-06,
      "loss": 1.0863,
      "step": 3880
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.425101214574899e-06,
      "loss": 1.0825,
      "step": 3890
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.421052631578948e-06,
      "loss": 1.0831,
      "step": 3900
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.417004048582996e-06,
      "loss": 1.0681,
      "step": 3910
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.412955465587044e-06,
      "loss": 1.0871,
      "step": 3920
    },
    {
      "epoch": 2.65,
      "learning_rate": 8.408906882591094e-06,
      "loss": 1.0531,
      "step": 3930
    },
    {
      "epoch": 2.66,
      "learning_rate": 8.404858299595142e-06,
      "loss": 1.1174,
      "step": 3940
    },
    {
      "epoch": 2.66,
      "learning_rate": 8.400809716599191e-06,
      "loss": 1.117,
      "step": 3950
    },
    {
      "epoch": 2.67,
      "learning_rate": 8.39676113360324e-06,
      "loss": 1.0938,
      "step": 3960
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.392712550607287e-06,
      "loss": 1.0736,
      "step": 3970
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.388663967611337e-06,
      "loss": 1.0872,
      "step": 3980
    },
    {
      "epoch": 2.69,
      "learning_rate": 8.384615384615385e-06,
      "loss": 1.1,
      "step": 3990
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.380566801619433e-06,
      "loss": 1.1201,
      "step": 4000
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.376518218623482e-06,
      "loss": 1.0755,
      "step": 4010
    },
    {
      "epoch": 2.71,
      "learning_rate": 8.37246963562753e-06,
      "loss": 1.0917,
      "step": 4020
    },
    {
      "epoch": 2.72,
      "learning_rate": 8.368421052631578e-06,
      "loss": 1.0873,
      "step": 4030
    },
    {
      "epoch": 2.73,
      "learning_rate": 8.364372469635628e-06,
      "loss": 1.0858,
      "step": 4040
    },
    {
      "epoch": 2.73,
      "learning_rate": 8.360323886639676e-06,
      "loss": 1.0721,
      "step": 4050
    },
    {
      "epoch": 2.74,
      "learning_rate": 8.356275303643725e-06,
      "loss": 1.138,
      "step": 4060
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.352226720647773e-06,
      "loss": 1.1194,
      "step": 4070
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.348178137651823e-06,
      "loss": 1.1364,
      "step": 4080
    },
    {
      "epoch": 2.76,
      "learning_rate": 8.344129554655871e-06,
      "loss": 1.1201,
      "step": 4090
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.34008097165992e-06,
      "loss": 1.0585,
      "step": 4100
    },
    {
      "epoch": 2.77,
      "learning_rate": 8.336032388663968e-06,
      "loss": 1.0613,
      "step": 4110
    },
    {
      "epoch": 2.78,
      "learning_rate": 8.331983805668016e-06,
      "loss": 1.1091,
      "step": 4120
    },
    {
      "epoch": 2.79,
      "learning_rate": 8.327935222672066e-06,
      "loss": 1.1446,
      "step": 4130
    },
    {
      "epoch": 2.79,
      "learning_rate": 8.323886639676114e-06,
      "loss": 1.0413,
      "step": 4140
    },
    {
      "epoch": 2.8,
      "learning_rate": 8.319838056680162e-06,
      "loss": 1.1225,
      "step": 4150
    },
    {
      "epoch": 2.81,
      "learning_rate": 8.315789473684212e-06,
      "loss": 1.1121,
      "step": 4160
    },
    {
      "epoch": 2.81,
      "learning_rate": 8.31174089068826e-06,
      "loss": 1.0534,
      "step": 4170
    },
    {
      "epoch": 2.82,
      "learning_rate": 8.307692307692307e-06,
      "loss": 1.1097,
      "step": 4180
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.303643724696357e-06,
      "loss": 1.1147,
      "step": 4190
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.299595141700405e-06,
      "loss": 1.1009,
      "step": 4200
    },
    {
      "epoch": 2.84,
      "learning_rate": 8.295546558704453e-06,
      "loss": 1.0584,
      "step": 4210
    },
    {
      "epoch": 2.85,
      "learning_rate": 8.291497975708503e-06,
      "loss": 1.1045,
      "step": 4220
    },
    {
      "epoch": 2.85,
      "learning_rate": 8.28744939271255e-06,
      "loss": 1.0559,
      "step": 4230
    },
    {
      "epoch": 2.86,
      "learning_rate": 8.2834008097166e-06,
      "loss": 1.063,
      "step": 4240
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.279352226720648e-06,
      "loss": 1.1671,
      "step": 4250
    },
    {
      "epoch": 2.87,
      "learning_rate": 8.275303643724696e-06,
      "loss": 1.077,
      "step": 4260
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.271255060728746e-06,
      "loss": 1.1119,
      "step": 4270
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.267206477732794e-06,
      "loss": 1.0773,
      "step": 4280
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.263157894736843e-06,
      "loss": 1.0874,
      "step": 4290
    },
    {
      "epoch": 2.9,
      "learning_rate": 8.259109311740891e-06,
      "loss": 1.0793,
      "step": 4300
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.25506072874494e-06,
      "loss": 1.1011,
      "step": 4310
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.251012145748987e-06,
      "loss": 1.1232,
      "step": 4320
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.246963562753037e-06,
      "loss": 1.0953,
      "step": 4330
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.242914979757085e-06,
      "loss": 1.0841,
      "step": 4340
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.238866396761134e-06,
      "loss": 1.1036,
      "step": 4350
    },
    {
      "epoch": 2.94,
      "learning_rate": 8.234817813765182e-06,
      "loss": 1.1207,
      "step": 4360
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.230769230769232e-06,
      "loss": 1.1033,
      "step": 4370
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.22672064777328e-06,
      "loss": 1.0719,
      "step": 4380
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.22267206477733e-06,
      "loss": 1.0539,
      "step": 4390
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.218623481781377e-06,
      "loss": 1.0818,
      "step": 4400
    },
    {
      "epoch": 2.97,
      "learning_rate": 8.214574898785425e-06,
      "loss": 1.1149,
      "step": 4410
    },
    {
      "epoch": 2.98,
      "learning_rate": 8.210526315789475e-06,
      "loss": 1.0624,
      "step": 4420
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.206477732793523e-06,
      "loss": 1.1197,
      "step": 4430
    },
    {
      "epoch": 2.99,
      "learning_rate": 8.20242914979757e-06,
      "loss": 1.0567,
      "step": 4440
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.0876251459121704,
      "eval_runtime": 10.6357,
      "eval_samples_per_second": 103.425,
      "eval_steps_per_second": 12.975,
      "step": 4447
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.19838056680162e-06,
      "loss": 1.0768,
      "step": 4450
    },
    {
      "epoch": 3.01,
      "learning_rate": 8.194331983805668e-06,
      "loss": 1.0687,
      "step": 4460
    },
    {
      "epoch": 3.02,
      "learning_rate": 8.190283400809716e-06,
      "loss": 1.1029,
      "step": 4470
    },
    {
      "epoch": 3.02,
      "learning_rate": 8.186234817813766e-06,
      "loss": 1.0569,
      "step": 4480
    },
    {
      "epoch": 3.03,
      "learning_rate": 8.182186234817814e-06,
      "loss": 1.0151,
      "step": 4490
    },
    {
      "epoch": 3.04,
      "learning_rate": 8.178137651821862e-06,
      "loss": 1.0405,
      "step": 4500
    },
    {
      "epoch": 3.04,
      "learning_rate": 8.174089068825911e-06,
      "loss": 1.07,
      "step": 4510
    },
    {
      "epoch": 3.05,
      "learning_rate": 8.17004048582996e-06,
      "loss": 1.0643,
      "step": 4520
    },
    {
      "epoch": 3.06,
      "learning_rate": 8.165991902834009e-06,
      "loss": 1.0859,
      "step": 4530
    },
    {
      "epoch": 3.06,
      "learning_rate": 8.161943319838057e-06,
      "loss": 1.0384,
      "step": 4540
    },
    {
      "epoch": 3.07,
      "learning_rate": 8.157894736842105e-06,
      "loss": 1.0873,
      "step": 4550
    },
    {
      "epoch": 3.08,
      "learning_rate": 8.153846153846154e-06,
      "loss": 1.0847,
      "step": 4560
    },
    {
      "epoch": 3.08,
      "learning_rate": 8.149797570850202e-06,
      "loss": 1.0105,
      "step": 4570
    },
    {
      "epoch": 3.09,
      "learning_rate": 8.145748987854252e-06,
      "loss": 1.0407,
      "step": 4580
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.1417004048583e-06,
      "loss": 1.0437,
      "step": 4590
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.13765182186235e-06,
      "loss": 1.0641,
      "step": 4600
    },
    {
      "epoch": 3.11,
      "learning_rate": 8.133603238866397e-06,
      "loss": 1.0662,
      "step": 4610
    },
    {
      "epoch": 3.12,
      "learning_rate": 8.129554655870445e-06,
      "loss": 1.0608,
      "step": 4620
    },
    {
      "epoch": 3.12,
      "learning_rate": 8.125506072874495e-06,
      "loss": 1.0728,
      "step": 4630
    },
    {
      "epoch": 3.13,
      "learning_rate": 8.121457489878543e-06,
      "loss": 1.039,
      "step": 4640
    },
    {
      "epoch": 3.14,
      "learning_rate": 8.117408906882591e-06,
      "loss": 1.0898,
      "step": 4650
    },
    {
      "epoch": 3.14,
      "learning_rate": 8.11336032388664e-06,
      "loss": 1.107,
      "step": 4660
    },
    {
      "epoch": 3.15,
      "learning_rate": 8.109311740890688e-06,
      "loss": 1.1001,
      "step": 4670
    },
    {
      "epoch": 3.16,
      "learning_rate": 8.105263157894738e-06,
      "loss": 1.0371,
      "step": 4680
    },
    {
      "epoch": 3.16,
      "learning_rate": 8.101214574898786e-06,
      "loss": 1.0945,
      "step": 4690
    },
    {
      "epoch": 3.17,
      "learning_rate": 8.097165991902834e-06,
      "loss": 1.0509,
      "step": 4700
    },
    {
      "epoch": 3.18,
      "learning_rate": 8.093117408906884e-06,
      "loss": 1.0694,
      "step": 4710
    },
    {
      "epoch": 3.18,
      "learning_rate": 8.089068825910931e-06,
      "loss": 1.0153,
      "step": 4720
    },
    {
      "epoch": 3.19,
      "learning_rate": 8.08502024291498e-06,
      "loss": 1.0373,
      "step": 4730
    },
    {
      "epoch": 3.2,
      "learning_rate": 8.080971659919029e-06,
      "loss": 1.0583,
      "step": 4740
    },
    {
      "epoch": 3.2,
      "learning_rate": 8.076923076923077e-06,
      "loss": 1.0213,
      "step": 4750
    },
    {
      "epoch": 3.21,
      "learning_rate": 8.072874493927125e-06,
      "loss": 1.0707,
      "step": 4760
    },
    {
      "epoch": 3.22,
      "learning_rate": 8.068825910931175e-06,
      "loss": 1.0801,
      "step": 4770
    },
    {
      "epoch": 3.22,
      "learning_rate": 8.064777327935222e-06,
      "loss": 1.0336,
      "step": 4780
    },
    {
      "epoch": 3.23,
      "learning_rate": 8.060728744939272e-06,
      "loss": 1.0829,
      "step": 4790
    },
    {
      "epoch": 3.24,
      "learning_rate": 8.05668016194332e-06,
      "loss": 1.0853,
      "step": 4800
    },
    {
      "epoch": 3.24,
      "learning_rate": 8.05263157894737e-06,
      "loss": 1.074,
      "step": 4810
    },
    {
      "epoch": 3.25,
      "learning_rate": 8.048582995951418e-06,
      "loss": 1.0805,
      "step": 4820
    },
    {
      "epoch": 3.26,
      "learning_rate": 8.044534412955467e-06,
      "loss": 1.0644,
      "step": 4830
    },
    {
      "epoch": 3.26,
      "learning_rate": 8.040485829959514e-06,
      "loss": 1.039,
      "step": 4840
    },
    {
      "epoch": 3.27,
      "learning_rate": 8.036437246963563e-06,
      "loss": 1.0326,
      "step": 4850
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.032388663967611e-06,
      "loss": 1.0776,
      "step": 4860
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.02834008097166e-06,
      "loss": 1.043,
      "step": 4870
    },
    {
      "epoch": 3.29,
      "learning_rate": 8.024291497975709e-06,
      "loss": 1.0784,
      "step": 4880
    },
    {
      "epoch": 3.3,
      "learning_rate": 8.020242914979758e-06,
      "loss": 1.0783,
      "step": 4890
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.016194331983806e-06,
      "loss": 1.0987,
      "step": 4900
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.012145748987854e-06,
      "loss": 1.0515,
      "step": 4910
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.008097165991904e-06,
      "loss": 1.0667,
      "step": 4920
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.004048582995952e-06,
      "loss": 1.1095,
      "step": 4930
    },
    {
      "epoch": 3.33,
      "learning_rate": 8e-06,
      "loss": 1.0245,
      "step": 4940
    },
    {
      "epoch": 3.34,
      "learning_rate": 7.99595141700405e-06,
      "loss": 1.0753,
      "step": 4950
    },
    {
      "epoch": 3.35,
      "learning_rate": 7.991902834008097e-06,
      "loss": 1.0522,
      "step": 4960
    },
    {
      "epoch": 3.35,
      "learning_rate": 7.987854251012147e-06,
      "loss": 1.0106,
      "step": 4970
    },
    {
      "epoch": 3.36,
      "learning_rate": 7.983805668016195e-06,
      "loss": 1.0948,
      "step": 4980
    },
    {
      "epoch": 3.37,
      "learning_rate": 7.979757085020243e-06,
      "loss": 1.0649,
      "step": 4990
    },
    {
      "epoch": 3.37,
      "learning_rate": 7.975708502024292e-06,
      "loss": 1.0096,
      "step": 5000
    },
    {
      "epoch": 3.38,
      "learning_rate": 7.97165991902834e-06,
      "loss": 1.0076,
      "step": 5010
    },
    {
      "epoch": 3.39,
      "learning_rate": 7.967611336032388e-06,
      "loss": 1.0605,
      "step": 5020
    },
    {
      "epoch": 3.39,
      "learning_rate": 7.963562753036438e-06,
      "loss": 1.0544,
      "step": 5030
    },
    {
      "epoch": 3.4,
      "learning_rate": 7.959514170040486e-06,
      "loss": 1.0638,
      "step": 5040
    },
    {
      "epoch": 3.41,
      "learning_rate": 7.955465587044534e-06,
      "loss": 1.021,
      "step": 5050
    },
    {
      "epoch": 3.41,
      "learning_rate": 7.951417004048583e-06,
      "loss": 1.0622,
      "step": 5060
    },
    {
      "epoch": 3.42,
      "learning_rate": 7.947368421052631e-06,
      "loss": 1.0452,
      "step": 5070
    },
    {
      "epoch": 3.43,
      "learning_rate": 7.943319838056681e-06,
      "loss": 1.0678,
      "step": 5080
    },
    {
      "epoch": 3.43,
      "learning_rate": 7.939271255060729e-06,
      "loss": 0.9841,
      "step": 5090
    },
    {
      "epoch": 3.44,
      "learning_rate": 7.935222672064778e-06,
      "loss": 1.0684,
      "step": 5100
    },
    {
      "epoch": 3.45,
      "learning_rate": 7.931174089068826e-06,
      "loss": 1.0393,
      "step": 5110
    },
    {
      "epoch": 3.45,
      "learning_rate": 7.927125506072876e-06,
      "loss": 1.0816,
      "step": 5120
    },
    {
      "epoch": 3.46,
      "learning_rate": 7.923076923076922e-06,
      "loss": 1.0744,
      "step": 5130
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.919028340080972e-06,
      "loss": 1.0575,
      "step": 5140
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.914979757085022e-06,
      "loss": 1.119,
      "step": 5150
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.91093117408907e-06,
      "loss": 1.0652,
      "step": 5160
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.906882591093117e-06,
      "loss": 0.9888,
      "step": 5170
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.902834008097167e-06,
      "loss": 1.0415,
      "step": 5180
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.898785425101215e-06,
      "loss": 1.0266,
      "step": 5190
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.894736842105263e-06,
      "loss": 1.0204,
      "step": 5200
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.890688259109313e-06,
      "loss": 1.0709,
      "step": 5210
    },
    {
      "epoch": 3.52,
      "learning_rate": 7.88663967611336e-06,
      "loss": 1.0342,
      "step": 5220
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.882591093117408e-06,
      "loss": 1.0492,
      "step": 5230
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.878542510121458e-06,
      "loss": 1.0375,
      "step": 5240
    },
    {
      "epoch": 3.54,
      "learning_rate": 7.874493927125506e-06,
      "loss": 1.0188,
      "step": 5250
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.870445344129556e-06,
      "loss": 1.0951,
      "step": 5260
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.866396761133604e-06,
      "loss": 1.0133,
      "step": 5270
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.862348178137651e-06,
      "loss": 1.0414,
      "step": 5280
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.858299595141701e-06,
      "loss": 1.0863,
      "step": 5290
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.854251012145749e-06,
      "loss": 1.0998,
      "step": 5300
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.850202429149797e-06,
      "loss": 1.0854,
      "step": 5310
    },
    {
      "epoch": 3.59,
      "learning_rate": 7.846153846153847e-06,
      "loss": 1.0594,
      "step": 5320
    },
    {
      "epoch": 3.6,
      "learning_rate": 7.842105263157896e-06,
      "loss": 1.0225,
      "step": 5330
    },
    {
      "epoch": 3.6,
      "learning_rate": 7.838056680161942e-06,
      "loss": 1.0728,
      "step": 5340
    },
    {
      "epoch": 3.61,
      "learning_rate": 7.834008097165992e-06,
      "loss": 1.0417,
      "step": 5350
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.82995951417004e-06,
      "loss": 1.0226,
      "step": 5360
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.82591093117409e-06,
      "loss": 1.0953,
      "step": 5370
    },
    {
      "epoch": 3.63,
      "learning_rate": 7.821862348178138e-06,
      "loss": 1.053,
      "step": 5380
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.817813765182187e-06,
      "loss": 1.0523,
      "step": 5390
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.813765182186235e-06,
      "loss": 1.1017,
      "step": 5400
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.809716599190285e-06,
      "loss": 1.0032,
      "step": 5410
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.805668016194333e-06,
      "loss": 1.0331,
      "step": 5420
    },
    {
      "epoch": 3.66,
      "learning_rate": 7.80161943319838e-06,
      "loss": 1.0483,
      "step": 5430
    },
    {
      "epoch": 3.67,
      "learning_rate": 7.79757085020243e-06,
      "loss": 1.0291,
      "step": 5440
    },
    {
      "epoch": 3.68,
      "learning_rate": 7.793522267206478e-06,
      "loss": 1.0113,
      "step": 5450
    },
    {
      "epoch": 3.68,
      "learning_rate": 7.789473684210526e-06,
      "loss": 1.0407,
      "step": 5460
    },
    {
      "epoch": 3.69,
      "learning_rate": 7.785425101214576e-06,
      "loss": 1.077,
      "step": 5470
    },
    {
      "epoch": 3.7,
      "learning_rate": 7.781376518218624e-06,
      "loss": 1.1063,
      "step": 5480
    },
    {
      "epoch": 3.7,
      "learning_rate": 7.777327935222672e-06,
      "loss": 1.0377,
      "step": 5490
    },
    {
      "epoch": 3.71,
      "learning_rate": 7.773279352226721e-06,
      "loss": 1.0841,
      "step": 5500
    },
    {
      "epoch": 3.72,
      "learning_rate": 7.76923076923077e-06,
      "loss": 1.0638,
      "step": 5510
    },
    {
      "epoch": 3.72,
      "learning_rate": 7.765182186234819e-06,
      "loss": 1.0525,
      "step": 5520
    },
    {
      "epoch": 3.73,
      "learning_rate": 7.761133603238867e-06,
      "loss": 1.0638,
      "step": 5530
    },
    {
      "epoch": 3.74,
      "learning_rate": 7.757085020242915e-06,
      "loss": 1.0478,
      "step": 5540
    },
    {
      "epoch": 3.74,
      "learning_rate": 7.753036437246964e-06,
      "loss": 1.0429,
      "step": 5550
    },
    {
      "epoch": 3.75,
      "learning_rate": 7.748987854251014e-06,
      "loss": 1.0615,
      "step": 5560
    },
    {
      "epoch": 3.76,
      "learning_rate": 7.74493927125506e-06,
      "loss": 1.0653,
      "step": 5570
    },
    {
      "epoch": 3.76,
      "learning_rate": 7.74089068825911e-06,
      "loss": 1.0755,
      "step": 5580
    },
    {
      "epoch": 3.77,
      "learning_rate": 7.736842105263158e-06,
      "loss": 1.0447,
      "step": 5590
    },
    {
      "epoch": 3.78,
      "learning_rate": 7.732793522267207e-06,
      "loss": 1.0372,
      "step": 5600
    },
    {
      "epoch": 3.78,
      "learning_rate": 7.728744939271255e-06,
      "loss": 1.1158,
      "step": 5610
    },
    {
      "epoch": 3.79,
      "learning_rate": 7.724696356275305e-06,
      "loss": 1.1195,
      "step": 5620
    },
    {
      "epoch": 3.8,
      "learning_rate": 7.720647773279351e-06,
      "loss": 1.0246,
      "step": 5630
    },
    {
      "epoch": 3.8,
      "learning_rate": 7.716599190283401e-06,
      "loss": 1.0339,
      "step": 5640
    },
    {
      "epoch": 3.81,
      "learning_rate": 7.712550607287449e-06,
      "loss": 1.0578,
      "step": 5650
    },
    {
      "epoch": 3.82,
      "learning_rate": 7.708502024291498e-06,
      "loss": 1.0747,
      "step": 5660
    },
    {
      "epoch": 3.82,
      "learning_rate": 7.704453441295546e-06,
      "loss": 1.0301,
      "step": 5670
    },
    {
      "epoch": 3.83,
      "learning_rate": 7.700404858299596e-06,
      "loss": 0.9873,
      "step": 5680
    },
    {
      "epoch": 3.84,
      "learning_rate": 7.696356275303644e-06,
      "loss": 1.0217,
      "step": 5690
    },
    {
      "epoch": 3.84,
      "learning_rate": 7.692307692307694e-06,
      "loss": 1.0941,
      "step": 5700
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.688259109311742e-06,
      "loss": 1.0842,
      "step": 5710
    },
    {
      "epoch": 3.86,
      "learning_rate": 7.68421052631579e-06,
      "loss": 1.0163,
      "step": 5720
    },
    {
      "epoch": 3.87,
      "learning_rate": 7.680161943319839e-06,
      "loss": 1.0387,
      "step": 5730
    },
    {
      "epoch": 3.87,
      "learning_rate": 7.676113360323887e-06,
      "loss": 1.0719,
      "step": 5740
    },
    {
      "epoch": 3.88,
      "learning_rate": 7.672064777327935e-06,
      "loss": 1.0495,
      "step": 5750
    },
    {
      "epoch": 3.89,
      "learning_rate": 7.668016194331985e-06,
      "loss": 1.0669,
      "step": 5760
    },
    {
      "epoch": 3.89,
      "learning_rate": 7.663967611336033e-06,
      "loss": 1.1226,
      "step": 5770
    },
    {
      "epoch": 3.9,
      "learning_rate": 7.65991902834008e-06,
      "loss": 1.0147,
      "step": 5780
    },
    {
      "epoch": 3.91,
      "learning_rate": 7.65587044534413e-06,
      "loss": 1.0244,
      "step": 5790
    },
    {
      "epoch": 3.91,
      "learning_rate": 7.651821862348178e-06,
      "loss": 1.0134,
      "step": 5800
    },
    {
      "epoch": 3.92,
      "learning_rate": 7.647773279352228e-06,
      "loss": 1.0215,
      "step": 5810
    },
    {
      "epoch": 3.93,
      "learning_rate": 7.643724696356276e-06,
      "loss": 1.1177,
      "step": 5820
    },
    {
      "epoch": 3.93,
      "learning_rate": 7.639676113360325e-06,
      "loss": 1.0472,
      "step": 5830
    },
    {
      "epoch": 3.94,
      "learning_rate": 7.635627530364373e-06,
      "loss": 1.0335,
      "step": 5840
    },
    {
      "epoch": 3.95,
      "learning_rate": 7.631578947368423e-06,
      "loss": 1.0346,
      "step": 5850
    },
    {
      "epoch": 3.95,
      "learning_rate": 7.627530364372468e-06,
      "loss": 1.0354,
      "step": 5860
    },
    {
      "epoch": 3.96,
      "learning_rate": 7.623481781376519e-06,
      "loss": 1.0663,
      "step": 5870
    },
    {
      "epoch": 3.97,
      "learning_rate": 7.619433198380567e-06,
      "loss": 1.0602,
      "step": 5880
    },
    {
      "epoch": 3.97,
      "learning_rate": 7.615384615384615e-06,
      "loss": 1.0458,
      "step": 5890
    },
    {
      "epoch": 3.98,
      "learning_rate": 7.611336032388663e-06,
      "loss": 1.079,
      "step": 5900
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.607287449392713e-06,
      "loss": 1.0888,
      "step": 5910
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.60323886639676e-06,
      "loss": 1.0395,
      "step": 5920
    },
    {
      "epoch": 4.0,
      "learning_rate": 7.5991902834008105e-06,
      "loss": 1.1178,
      "step": 5930
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.0923843383789062,
      "eval_runtime": 10.6309,
      "eval_samples_per_second": 103.472,
      "eval_steps_per_second": 12.981,
      "step": 5930
    }
  ],
  "max_steps": 22230,
  "num_train_epochs": 15,
  "total_flos": 6.084121621168128e+18,
  "trial_name": null,
  "trial_params": null
}